Downloaded from rspb.royalsocietypublishing.org on March 2, 2010

Evidence that female preferences have shaped male signal evolution in a clade of specialized plant-feeding insects

Rafael L Rodríguez, Karthik Ramaswamy and Reginald B Cocroft

Proc. R. Soc. B 2006 273, 2585-2593 doi: 10.1098/rspb.2006.3635

References This article cites 43 articles, 5 of which can be accessed free http://rspb.royalsocietypublishing.org/content/273/1601/2585.full.html#ref-list-1 Receive free email alerts when new articles cite this article - sign up in the box at the top Email alerting service right-hand corner of the article or click here

To subscribe to Proc. R. Soc. B go to: http://rspb.royalsocietypublishing.org/subscriptions

This journal is © 2006 The Royal Society Downloaded from rspb.royalsocietypublishing.org on March 2, 2010

Proc. R. Soc. B (2006) 273, 2585–2593 doi:10.1098/rspb.2006.3635 Published online 1 August 2006

Evidence that female preferences have shaped male signal evolution in a clade of specialized plant-feeding insects Rafael L. Rodrı´guez*, Karthik Ramaswamy and Reginald B. Cocroft 223 Tucker Hall, Biological Sciences, University of Missouri–Columbia, Columbia, MO 65211-7400, USA Mate choice is considered an important influence in the evolution of mating signals and other sexual traits, and—since divergence in sexual traits causes reproductive isolation—it can be an agent of population divergence. The importance of mate choice in signal evolution can be evaluated by comparing male signal traits with female preference functions, taking into account the shape and strength of preferences. Specifically, when preferences are closed (favouring intermediate values), there should be a correlation between the preferred values and the trait means, and stronger preferences should be associated with greater preference–signal correspondence and lower signal variability. When preferences are open (favouring extreme values), signal traits are not only expected to be more variable, but should also be shifted towards the preferred values. We tested the role of female preferences in signal evolution in the Enchenopa binotata species complex of treehoppers, a clade of plant-feeding insects hypothesized to have speciated in sympatry. We found the expected relationship between signals and preferences, implicating mate choice as an agent of signal evolution. Because differences in sexual communication systems lead to reproductive isolation, the factors that promote divergence in female preferences—and, consequently, in male signals—may have an important role in the process of speciation. Keywords: vibrational signalling; sexual coevolution; phytophagous insects; Hemiptera; Membracidae

1. INTRODUCTION influence the variability of signal traits; (iv) when Mate choice is widespread in nature and regarded as comparing closed preferences with open preferences, a strong and pervasive influence in the evolution of which favour values in one extreme of the range, the mating signals and other sexual traits (Darwin 1871; former will be associated with lower signal variability West-Eberhard 1983; Andersson 1994). The importance (Gerhardt 1994; Shaw & Herlihy 2000; Gerhardt & Huber of mate choice in signal evolution can be evaluated by 2002); and (v) for open preferences that differ in strength comparing signal traits with mate preferences. Prefer- or sign, signal traits should be shifted towards the ences can be described using curves that plot attractive- preferred values (Houde & Endler 1990; Endler & ness as a function of variation in signal traits, and Houde 1995; Simmons et al. 2001). characterizing preference functions generates hypotheses We evaluated the importance of female mate choice in about the type of selection they exert on signals (Butlin signal evolution by testing these predictions across four 1993; Gerhardt 1994; Ritchie 1996; Jang & Greenfield species in the Enchenopa binotata species complex of 1998; Gerhardt et al. 2000; Shaw & Herlihy 2000; treehoppers. This complex consists of 11 or more species Ritchie et al. 2001; Parri et al. 2002; Blows et al. 2003; that occur in close sympatry across eastern North America, and differ in male mating signals. In the mating system Klappert & Reinhold 2003; Brooks et al. 2005; Gerhardt of these insects, male–female duetting facilitates pair 2005a,b; Bentsen et al. 2006). Comparison of prefer- formation, so female response signals provide an assay of ences with the distribution of signal traits for closely mate preference (Rodrı´guez et al. 2004). We found that related species or populations can reveal the extent to preferences differ among species in the E. binotata which signals have responded to selection exerted complex, and that signals match the values preferred by by preferences. the females to a degree that is determined by preference The hypothesis that mate choice is important in signal strength and shape. evolution makes five predictions about the preference– signal relationship: (i) for closed preferences—which favour intermediate over extreme values—mean signal values will correspond to the values preferred by females 2. MATERIAL AND METHODS (Ritchie 1996; Shaw 2000; Mendelson & Shaw 2002; (a) The Enchenopa binotata species complex Gerhardt 2005a,b); (ii) preference strength will influence (Hemiptera: Membracidae) the degree of correspondence between preferred values Each species in this clade specializes on a different species of woody plant in the forest understory, edge and canopy. The and mean signal traits; (iii) preference strength will E. binotata complex is a case study of sympatric speciation involving shifts to novel host plants (Wood & Guttman 1983; * Author for correspondence ([email protected]). Wood & Keese 1990; Wood et al. 1990; Lin & Wood 2002;

Received 21 April 2006 2585 q 2006 The Royal Society Accepted 29 May 2006 Downloaded from rspb.royalsocietypublishing.org on March 2, 2010

2586 R. L. Rodrı´guez and others Female preferences and signal evolution

(a) included dominant frequency at the end of the whine, whine length, number of signals per bout, length of the interval 1000 between signals, pulse number and pulse rate (figure 1). We varied one trait at a time while keeping the other traits set to the mean of the local population from which females were drawn. The values assayed for each trait span the range of

frequency (Hz) frequency 200 variation in the complex. For signal frequency, this range 200 ms (100–500 Hz) did not reach frequencies high enough to describe the full shape of the E. binotata ‘Celastrus’ preference, so we extended the range of frequencies to whine pulses 720 Hz for two additional females. (b) Rather than relying on natural variation in mating signals, we synthesized signals with a custom-written program in MATLAB v. 5.2.1 (Mathworks Inc., Natick, 1 s MA). Preliminary experiments showed that these signals are as effective in eliciting female responses as Figure 1. (a) Spectrogram and waveform of a recording of a playbacks of recorded male signals (R. L. Rodrı´guez 2006, signal produced by a male E. binotata ‘Viburnum’. (b) Bout consisting of four signals that increase in amplitude. unpublished data). The artificially generated signals use a constant frequency rather than a frequency sweep; this Cocroft et al. in press). Host shifts cause assortative mating, greatly streamlines the playback procedure, as they do not because the insects’ life-history timing is regulated by host require compensation for differential frequency filtering plant phenology, so insects on plants with different phenol- during propagation along plant substrates (Cocroft & ogies acquire allochronic life cycles. Host shifts also result in Rodrı´guez 2005). divergent ecological selection promoting host specialization Stimuli had the typical E. binotata signal bout structure and host fidelity. of gradually increasing amplitude (figure 1b). We used a Sexual communication is important for assortative mating structure of four signals that increased in amplitude in in this complex. As in many herbivorous insects (Claridge the pattern: 25, 75, 100 and 100% of maximum amplitude. 1985; Henry 1994; Cokl & Virant-Doberlet 2003; Cocroft & We varied this pattern only for the experiment testing the Rodrı´guez 2005), pair formation in E. binotata is facilitated signal number, where the one-signal stimulus had 100% by exchanges of plant-borne vibrational signals between amplitude and the two-signal stimulus had a pattern of 75 males and females (Hunt 1994; Sattman & Cocroft 2003; and 100% amplitude. Rodrı´guez et al. 2004). Mate-searching males produce signals, and female responses elicit localized searching by (d) Stimulation and recording of female responses the males. Signal structure is conserved across the complex, We played back stimuli to females through the stems of and consists of a tone with harmonics and a frequency sweep, potted plants (ca 50 cm tall) of their host species. To the whine, followed by a series of pulses (figure 1a). In spite introduce stimuli to a plant stem, we attached a magnet to of overall similarity in signal structure, there is quantitative the stem with wax (Endevco, San Juan Capistrano, CA) signal variation among species and individuals (R. B. Cocroft and placed an electromagnet ca 2mmfromit.The 2006, unpublished data). Females choose among males by electromagnet received signals from a Macintosh G4 responding to signals on the basis of this variation, computer, amplified with an Optimus MPA-40 amplifier. influencing the likelihood of being located by males Stimulus presentation was controlled with a custom-written (Rodrı´guez et al. 2004). program in MATLAB. The plant was placed on a vibration isolation table (Vibraplane, Kinetic Systems, Boston, MA) (b) General to minimize the noise generated by building vibrations. We Experiments were performed during April–August isolated the plant from the table with shock-absorbing 2003–2005. To ensure that females were sexually receptive sorbothane (Edmund Scientifics, Tonawanda, NY), further and responsive to signals, we tested virgin females reared from to reduce building vibrations and isolate the plant from nymphs collected in Boone County, MO. We reared nymphs table resonance. on their host plants in an outdoor facility in the University of For testing, we placed females on the prepared stems, ca Missouri–Columbia campus. We tested females 4–6 weeks 5 cm from the magnet. We recorded the playbacks and the after their adult moult, at the peak of their receptivity (Wood elicited female responses by focusing the beam of a laser 1993). We tested females from the E. binotata species that vibrometer (Polytec CLV 1000 with a CLV M030 decoder specialize on Celastrus scandens, Cercis canadensis, Ptelea module; Polytec Inc., Auburn, MA) on a small (ca 2mm2) trifoliata and Viburnum rufidulum host plants. The male piece of reflective tape affixed to the stem. The laser beam signals of these species offer a representative example of was approximately perpendicular to the stem, and the laser variation in the complex, including its extremes in frequency source was ca 50 cm from the plant. The laser signal was and length, and two intermediate species. Species in this high-pass filtered at 60 Hz (Krohn-Hite 3202; Krohn-Hite complex await formal description, and we refer to them by the Corporation, Brockton, MA). The output of the filter was name of their host plant, thus E. binotata ‘Celastrus’, etc. sent to a Macintosh G4 computer via an Edirol UA-5 USB interface (Roland Corporation, Japan), and recorded with (c) Stimulus design SOUNDEDIT 16 v. 2 (Macromedia Inc., San Francisco, CA) We tested the effect of variation in six signal traits that differ at a sampling rate of 44.1 kHz. We monitored the playbacks among species in the E. binotata complex (Rodrı´guez et al. and female responses using a Radio Shack MPA-45 2004; R. B. Cocroft 2006, unpublished data). These traits amplifier connected to an RCA loudspeaker and a Hameg

Proc. R. Soc. B (2006) Downloaded from rspb.royalsocietypublishing.org on March 2, 2010

Female preferences and signal evolution R. L. Rodrı´guez and others 2587

, , , E. binotata Cercis, E. binotata Viburnum, E. binotata , Ptelea, E. binotata Celastrus,

1.0

0.5

0 100 200 300 400 500 100 200 300 400 500 100 200 300 400 500 100 300 500 700 frequency (Hz)

1.0

0.5

0 100 500 900 1300 100 500 900 1300 100 500 900 1300 100 500 900 1300 whine length (ms)

1.0

0.5

0 0 4 8 12 16 0 4 8 12 16 0 4 8 12 16 0 4 8 12 16 signal number

1.0

0.5

0

proportion of females responding to stimulus 1000 3000 5000 1000 3000 5000 1000 3000 5000 1000 3000 5000 signal interval (ms)

1.0

0.5

0 04812048120481204812 pulse number

1.0

0.5

0 6 10 14 18 22 26 6 10 14 18 22 26 6 10 14 18 22 26 6 10 14 18 22 26 pulse rate (Hz) Figure 2. Proportion of females responding to stimuli varying in frequency, whine length, signal number, signal interval, pulse number and pulse rate. Each column corresponds to one species. Dots indicate the proportion of females that responded; lines indicate the spline regressions of those proportions G1 s.e. The x-axes show the range of variation tested for each signal trait, which corresponds to the range in the E. binotata species complex.

HM 203-7 20 MHz oscilloscope (Hameg Instruments, ca 5 cm away, on the basis of measurements of over 10 Mainhausen, Germany). Air temperature was kept at males from three species in the E. binotata complex (from 24–25 8C. Cercis, Ptelea and Viburnum; R. L. Rodrı´guez and G. McNett We adjusted stimulus peak amplitude at the point where 2006, personal observation). Females usually walked a short the females were placed on the plant stem (ca 5cm distance after being placed on the stem, so they received from the magnet). We used an amplitude (0.3 mm sK1) stimulation within a few centimetres from the point of corresponding to the average amplitude of a male signalling amplitude calibration.

Proc. R. Soc. B (2006) Downloaded from rspb.royalsocietypublishing.org on March 2, 2010

2588 R. L. Rodrı´guez and others Female preferences and signal evolution

Table 1. Analysis of the responses of E. binotata females, testing the effect of variation in stimulus traits, species, their interaction and female individual identity.

factor d.f. c2 p frequency species 3 0 1.0 linear 1 41.23 !0.0001 species!linear 3 176.28 !0.0001 quadratic 1 3.77 0.05 species!quadratic 3 154.18 !0.0001 individual 48 123.67 !0.0001 whine length species 3 0 1 linear 1 0.22 0.64 species!linear 3 31.47 !0.0001 quadratic 1 15.1 0.0001 species!quadratic 3 10.74 0.013 individual 44 255.22 !0.0001 signal number species 3 0.0001 1 linear 1 12.64 0.0004 species!linear 3 5.41 0.14 quadratic 1 0.003 0.96 species!quadratic 3 0.18 0.98 individual 49 184.43 !0.0001 signal interval species 3 K0.00003 — linear 1 0.0000003 1 species!linear 3 2.71 0.44 quadratic 1 0.000003 1 species!quadratic 3 2.89 0.089 individual 40 174.68 !0.0001 pulse number species 3 0 1.0 linear 1 4.99 0.025 species!linear 3 9.65 0.022 quadratic 1 2.67 0.10 species!quadratic 3 3.97 0.26 individual 41 145.29 !0.0001 pulse rate species 3 0.0002 1 linear 1 0.13 0.72 species!linear 3 15.49 0.0014 quadratic 1 2.96 0.085 species!quadratic 3 3.83 0.28 individual 47 239.75 !0.0001

(e) Female preference analysis using the interaction between stimulus and species as an Our experimental unit is an individual female; the data indicator (see Olvido & Wagner 2004). Lack of interaction contributed by each female are her responses to variation in a indicates preferences that are similar among species. signal trait. In each of the experiments testing the effect of Because linear and quadratic terms can yield a simplistic variation in a signal trait, we presented each female with a view of the actual shape of preferences (Brodie et al. 1995; random sequence of all the stimuli representing variation in Ritchie 1996; Blows et al. 2003; Brooks et al. 2005), we used that trait. The time between each stimulus was 15 s, and the cubic splines to characterize preferences without making time between experiments testing the effect of variation in assumptions about their shape (Schluter 1988). To this end, different signal traits was 2 min. The number of females we scaled to 1 the proportion of females responding to the tested in each experiment was 13–15 for each of the three stimuli, dividing by the maximum proportion of females species, but for E. binotata ‘Cercis’ nZ7 for frequency, signal that responded in each experiment to each species. We G number and pulse rate; nZ4 for whine length; nZ3 for pulse used these data to calculate splines 1 s.e. based on number; and nZ1 for signal interval (the small sample for 100 bootstraps with the program created by D. Schluter w this species was owing to low survivorship among the nymphs (www.zoology.ubc.ca/ schluter/lab.html). This program we collected). calculates a running regression with a window size optimizing We observed whether females responded to one or more the prediction of deleted observations. This window signalsinastimulusbout,analysing the recordings determines the curve stiffness, with larger windows giving smoother functions. We used the squared standard deviation with SOUNDEDIT. We used logistic regressions (Hosmer & of the values calculated by the spline regressions as an index Lemeshow 2000) to analyse variation in female responses of preference strength, as in Schluter (1988). according to stimulus variable (tested as linear and quadratic terms), species and female individual identity (as a random term; see Rodrı´guez et al. 2004). We report the likelihood (f ) Preference–signal correspondence ratio c2-tests for these terms. The main purpose of this To compare preferences with the distribution of signal traits, analysis was to identify species differences in preferences, we obtained the distribution of male signal traits from

Proc. R. Soc. B (2006) Downloaded from rspb.royalsocietypublishing.org on March 2, 2010

Female preferences and signal evolution R. L. Rodrı´guez and others 2589

Table 2. Relationship between the preference strength and the distribution of signal traits in E. binotata. (The effect of preference strength on the signal–preference mismatch was tested only for closed preferences. The effect of preference strength on the CVof signal traits was tested separately for closed and open preferences.)

closed preferences open preferences

factor d.f. Fp factor d.f. Fp preference–signal mismatch strength 1, 2 14.21 0.064 species 3, 2 3.19 0.25 interaction 3, 2 0.80 0.60 male signal trait CV strength 1, 2 7.8 0.11 strength 1, 6 0.38 0.56 species 3, 2 0.93 0.56 species 3, 6 0.23 0.87 interaction 3, 2 0.22 0.88 interaction 3, 6 0.20 0.89 a library of recordings of the Missouri populations from among species lay in the preferred values (figures 2 and 3). which the females were drawn (R. B. Cocroft 2006, The standard error curves (figure 3) show that preferred unpublished data). Signals were recorded using laser frequencies differed for all four species. Preferred whine vibrometry (as described previously), with the laser beam lengths were similar for E. binotata ‘Cercis’ and E. binotata focused on a small piece of reflective tape within a few ‘Viburnum’, and for E. binotata ‘Ptelea’ and E. binotata centimetres of a male on the stem of a potted host plant. ‘Celastrus’, with females from the first two species Differences between the mean temperatures at which females preferring longer whines than females from the last two were tested and males recorded were less than 1 8C. species (figure 3). Preferences for signal number were We tested the relationship between preferences and signals open (favouring more signals in a bout) and similar among using the between-male coefficient of variation (CV), and the species (figure 2). Preferences relating to signal pulses preference–signal mismatch using the difference between the differed among species in shape: preferences for pulse peak of the preference and the mean of the signal trait (where number were either open (favouring more pulses), flat 0 is perfect correspondence). We estimated peaks (points of (no preference) or closed (figure 2), while preferences maximum response) only for closed preferences. If there was for pulse rate were either open (favouring slower more than one point at the peak, we took the mean of those rates in two species and faster rates in one species) or points. We then obtained the absolute value of the difference closed (figure 2). between the preference peak and the mean of the male signal Because we presented each female with a random trait, and standardized for among-species and among-trait sequence of all stimuli for any one signal trait, we were able comparisons by dividing it by the male mean. We used least- to test the effect of individual variation on how females squares regressions, including preference strength, species responded to the stimuli. We found a significant individual and their interaction in the model. Ongoing studies of component in all preferences (table 1), which is consistent variation in mitochondrial and nuclear genes within and with individual variation in preferences. among species in the E. binotata complex (R. L. Snyder 2006, unpublished data) reveal that they are very similar for the (b) Preference–signal correspondence genetic markers used (perhaps reflecting recent divergence), We compared preferences with the distribution of signal and their relationships are not yet resolved; accordingly, here traits among species (figure 3) to test the predictions of the we treat their relationships as a ‘star’ phylogeny, i.e. four hypothesis that female preferences have been important in closely related but independent lineages. shaping male signal evolution. There were two signal traits for which females from all four species had closed 3. RESULTS preferences—frequency and whine length (figure 3)—so (a) Female preferences we could test prediction (i), i.e. for closed preferences Female responses were influenced by variation in most of the mean signal values will correspond to the preferred values. signal traits tested; only signal interval had no influence There was a very close match between preferences and (figure 2; table 1). (For signal interval, the c2 was negative, signals, with the match being closer for frequency than for probably owing to a rounding error; the very low magnitude whine length (figure 4a(i), (ii)). of the term indicates no effect.) Note that for E. binotata To test prediction (ii), i.e. preference strength will ‘Cercis’, we were able to test only one female for signal influence the correspondence between preferred values interval; as with the other species, there was no indication of and signal traits, we used all closed preferences: frequency apreference(figure 2). The first step in comparing and whine length in the four species, pulse number in preferences among species was the test of the stimulus! E. binotata ‘Celastrus’, and pulse rate in E. binotata species interaction. For most preferences, this interaction ‘Viburnum’ (figure 3). We found a marginally significant was significant, either for the linear or quadratic terms, or relationship between preference strength and the both (table 1). Only for the signal number was there no preference–signal mismatch (table 2), but the magnitude interaction (table 1), indicating that preferences for signal of the F-ratio (equal to 14.21) shows that this is owing number were similar among species (figures 2 and 3). We to low statistical power rather than to a weak effect. then focused on the shape of the preference functions to Since the preference strength!species interaction was elucidate differences among species. non-significant and had a small F-ratio (table 2), we Preferences for signal frequency and whine length were removed it from the model to increase power. This test closed (favouring intermediate values), and differences revealed a significant relationship with preference strength

Proc. R. Soc. B (2006) Downloaded from rspb.royalsocietypublishing.org on March 2, 2010

2590 R. L. Rodrı´guez and others Female preferences and signal evolution

(F1,5Z16.44, pZ0.0098): stronger preferences were yields strong evidence that female choice has been an associated with smaller mismatches (figure 4a(iii)). important agent of signal evolution. Prediction (iii) states that preference strength will Members of the E. binotata complex specialize on influence signal trait variability. We tested this prediction different host plants, so one important issue is the extent separately for closed and open preferences, because the to which species differences in preferences are owing to former were stronger (figure 4b; table 3). For closed their development on different host plants, i.e. to preferences, the relationship between preference strength phenotypic plasticity and host plant effects. We have not and signal trait CVs was non-significant (table 2), but the evaluated plasticity in E. binotata female preferences, but magnitude of the F-ratio (equal to 7.8) shows that this is three lines of evidence suggest a minor role for plasticity in owing to low statistical power rather than a weak effect. Since preference divergence. First, plasticity has little influence there was no preference strength!species interaction on the other half of the signal-receiver system: inducing (table 2), we removed the interaction term from the model male E. binotata to signal on foreign host plants causes little to increase power. This test revealed a significant relation- or no change in most signal traits—except that, being ship between preference strength and signal trait CVs host-specific, males produce fewer and shorter signals (F1,5Z15.15, pZ0.012): stronger preferences were associ- (Sattman & Cocroft 2003). Rearing males on different host ated with lower CVs (figure 4b). By contrast, there was no plants reveals substantial plasticity and genotype! such relationship for open preferences (table 2; figure 4b), environment interactions, but plasticity does not exceed probably because they were much weaker (figure 4b(iii); the range of variation of the species (R. L. Rodrı´guez 2006, table 3), while the signal traits showed much more variation unpublished data). Second, studies offemale preferences in in their CVs (compare the axes of figure 4b(i) and (ii)). other species reveal plasticity and genotype!environment Prediction (iv) states that closed preferences will be interactions, but again these do not exceed the range of associated with lower signal variability than open pre- variation of the species (Rodrı´guez & Greenfield 2003). ferences. We found a significant relationship between Third, the host-specificity of the members of the E. binotata preference shape and signal trait CVs that met this complex (Wood & Guttman 1983) suggests that reaction prediction (figure 4c; table 3). norms are unlikely to have been shaped by selection on Prediction (v) states that, for open preferences, signal performance on multiple host plants, but instead by traits should be shifted towards the values preferred by the selection on single hosts. Thus, the observed preference– females. We could apply this prediction only to the signal correspondence would be unlikely to arise as a preference for pulse rate, where females from three species consequence of unselected variation in reaction norms had open preferences differing in sign. Females from in novel environments. We suggest that E. binotata E. binotata ‘Cercis’ and E. binotata ‘Ptelea’ preferred slower communication systems have probably been shaped by pulse rates, while females from E. binotata ‘Celastrus’ selection on each species’ host, with differences in preferred faster pulse rates (figure 2). Male pulse rates preferences arising from divergent selection among species. differed accordingly (figure 3). The other species, E. binotata Several factors may influence selection on female ‘Viburnum’, had a closed preference for pulse rate, with the preferences. Members of the E. binotata complex are preferred value matching that of the males (figure 3). highly host-specific and show reduced fitness on the host plants of other species in the complex (Wood & Guttman 1983). Although partial reproductive isolation arises from host fidelity and allochronic life histories owing to 4. DISCUSSION differences in host plant phenology (Wood & Keese Females showed preferences for most of the signal traits 1990; Wood et al. 1990), mating seasons may overlap, that vary among species in the E. binotata complex. Most and mate-searching males occasionally disperse across preferences differed among species, with females preferring host plants (R. B. Cocroft 2006, personal observation). different signal values and attending in varying ways and Thus, mating with heterospecifics may influence offspring degrees to different signal traits. These species differences fitness, and mate recognition may be important. Female in female preferences suggest that there may have been a choice may also influence offspring fitness in other ways. pattern of divergent selection on male signals in the There is substantial repeatability (Sattman & Cocroft E. binotata complex. We tested five predictions of this 2003) and genetic variation in several male signal traits hypothesis among four closely related species, finding (R. L. Rodrı´guez 2006, unpublished data); further, support for all of them: (i) for closed preferences, the mean individual variation in female responses to playbacks values of male signal traits corresponded to the values (Rodrı´guez et al. 2004; this study) is consistent with preferred by females; (ii) preference strength influenced the individual differences and genetic variation in preferences. degree of this correspondence; (iii) preference strength Genetic variation in preferences and signals will result influenced the variability of signal traits; (iv) closed in Fisherian coevolution, facilitating divergence (West- preferences were associated with lower variability in signal Eberhard 1983; Kokko et al. 2002; Mead & Arnold 2004). traits; and (v) open preferences differing in sign were In addition, variation in male signals may be condition- associated with the expected differences in signal traits. dependent (Jennions et al. 2001; Brandt & Greenfield Previous tests of the role of mate choice in signal evolution 2004; Hunt et al. 2004), and if correlations between signal have lent support to the first (Ritchie 1996; Shaw 2000; traits and condition vary among host plants, there may be Mendelson & Shaw 2002; Gerhardt 2005a,b), second divergent selection on preferences (Cocroft et al. in press). (Butlin 1993), fourth (Gerhardt 1994; Shaw & Herlihy Another factor that may influence preference divergence 2000; Gerhardt & Huber 2002) and fifth (Houde & Endler involves the interaction between sexual selection and 1990; Endler & Houde 1995) predictions. This is the first adaptation to novel environments, which is of particular study to test all these predictions comprehensively, and relevance in the context of speciation and host shifts in

Proc. R. Soc. B (2006) Downloaded from rspb.royalsocietypublishing.org on March 2, 2010

Female preferences and signal evolution R. L. Rodrı´guez and others 2591 t distribution male signal trai 1

0.5 preference 0 100 200 300 400 500 600 700 100 500 900 1 300 0 4 8 12 16 frequency (Hz) whine length (ms) signal number distribution male signal trait 1

0.5 preference 0 0 4 8 12 6 10 14 18 22 26 1 000 2 000 3 000 4 000 5 000 pulse number pulse rate (Hz) signal interval (ms) Figure 3. Comparison of preferences and signals. Curves indicate spline regressions as in figure 2, shown are the G1 s.e. curves for frequency and whine length to emphasize species differences in preference peaks. For other preferences, differences lay in the shape rather than in the peaks, so we only show the regression curves for simplicity. Histograms show the distribution of male traits. Species are indicated by colours: black, E. binotata ‘Cercis’; red, E. binotata ‘Viburnum’; green, E. binotata ‘Ptelea’; blue, E. binotata ‘Celastrus’.

(a) frequency whine length r = 0.99, p = 0.0004 r = 0.99, p = 0.0097 (i) (ii) (iii) 500 800 0.4

400 700 0.3 600 300 0.2 500 200 400 0.1 (c) 50 male mean ± Cl (ms) male mean ± Cl (Hz) 100 300 0 40 preference–signal mismatch 100 200 300 400 500 300 400 500 600 700 800 0 0.1 0.2 30 preference peak (Hz) preference peak (ms) preference strength 20 (b)(i)closed preferences (ii) open preferences (iii) 20 50 0.15 10 male signal trait CV 0 15 40 0.10 closed open 30 10 preference shape 20 0.05 5 10 preference strength male signal trait CV 0 male signal trait CV 0 0 0 0.1 0.20 0.005 0.01 closed open preference strength preference strength preference shape Figure 4. Preference–signal relationships. Species are indicated by colours as in figure 3.(a(i), (ii)) For frequency and whine length there was a strong correlation between preference peaks and mean signal values; dashed lines indicate a 1:1 relationship. (a(iii)) Stronger closed preferences were associated with smaller preference–signal mismatches. (b(i), (ii)) Preference, strength influenced signal trait CVs for closed but not for open preferences. (b(iii)) Closed preferences were stronger than open preferences. (c) Closed preferences were associated with lower signal trait CVs than open preferences.

Proc. R. Soc. B (2006) Downloaded from rspb.royalsocietypublishing.org on March 2, 2010

2592 R. L. Rodrı´guez and others Female preferences and signal evolution

Table 3. Relationship between the preference shape (closed versus open) and the coefficient of variation of male signal traits, and female preference strength in E. binotata.

relationship between preference shape and male signal relationship between preference shape and trait CV strength factor d.f. Fp d.f. Fp shape 1, 11 6.10 0.031 1, 11 6.30 0.029 species 3, 11 0.42 0.99 3, 11 0.14 0.93 interaction 3, 11 0.08 0.97 3, 11 0.09 0.97

plant-feeding insects (Cocroft et al. in press). Condition- Brooks, R., Hunt, J., Blows, M. W., Smith, M. J., Bussie`re, related mate choice may facilitate adaptation to novel L. F. & Jennions, M. D. 2005 Experimental evidence for environments (Lorch et al.2003) and, conversely, multivariate stabilizing sexual selection. Evolution 59, adaptation to novel host plants may result in divergence 871–880. (doi:10.1554/04-662) in preferences (Rundle & Nosil 2005). Finally, our finding Butlin, R. K. 1993 The variability of mating signals and of strong preference–signal correspondence suggests that preferences in the brown planthopper, Nilaparvata lugens (Homoptera: Delphacidae). J. Insect Behav. 6, 125–140. models of sexual conflict (Tregenza et al. 2006) are not (doi:10.1007/BF01051499) important in E. binotata mating systems. This idea is in Claridge, M. F. 1985 Acoustic signals in the Homoptera: accordance with the active role that duetting females play behavior, taxonomy, and evolution. Ann. Rev. Entomol. 30, in pair formation, suggesting that they are not exposed to 297–317. (doi:10.1146/annurev.en.30.010185.001501) male harassment (Alexander et al. 1997). Cocroft, R. B. & Rodrı´guez, R. L. 2005 The behavioral In conclusion, the E. binotata species complex offers a ecology of insect vibrational communication. BioScience strong example of female preferences having a key role in 55, 323–334. (doi:10.1641/0006-3568(2005)/0006-3568 signal evolution, where following a host shift sexual (2005)055[0323:TBEOIV]2.0.CO;2) communication may act in synergy with host fidelity, Cocroft, R. B., Rodrı´guez, R. L. & Hunt, R.E. In press. Host plasticity in life-history timing and ecological selection shifts, the evolution of communication and speciation in (Wood & Guttman 1983; Wood 1993; Cocroft et al. the Enchenopa binotata complex of treehoppers. In in press) in the process of speciation. Specialization, speciation, and radiation: the evolutionary biology of herbivorous insects (ed. K. Tilmon). Berkeley, CA: We thank Michael D. Greenfield, Gerlinde Ho¨bel and two University of California Press. anonymous reviewers for their comments on the manuscript, Cokl, A. & Virant-Doberlet, M. 2003 Communication with and Barb Sonderman for her logistic support. This research substrate-borne signals in small plant-dwelling insects. adhered to the Association for the Study of Animal Annu. Rev. Entomol. 48, 29–50. (doi:10.1146/annurev. Behaviour/Animal Behavior Society Guidelines for the Use ento.48.091801.112605) of Animals in Research, the legal requirements of the USA, Darwin, C. 1871 The descent of man and selection in relation to and all institutional guidelines. Financial support was sex. London, UK: J. Murray. provided by the National Science Foundation grant IBN Endler, J. A. & Houde, A. E. 1995 Geographic variation in 0318326 to R.B.C. and R.L.R. female preferences for male traits in Poecilia reticulata. Evolution 49, 456–468. (doi:10.2307/2410270) Gerhardt, H. C. 1994 The evolution of vocalization in frogs REFERENCES and toads. Annu. Rev. Ecol. Syst. 25, 293–324. (doi:10. Alexander, R. D., Marshall, D. C. & Cooley, J. R. 1997 1146/annurev.es.25.110194.001453) Evolutionary perspectives on insect mating. In The Gerhardt, H. C. 2005a Acoustic spectral preferences in two evolution of mating systems in insects and arachnids (ed. cryptic species of grey treefrogs: implications for mate J. C. Choe & B. J. Crespi), pp. 4–31. Cambridge, UK: choice and sensory mechanisms. Anim. Behav. 70, 39–48. Cambridge University Press. (doi:10.1016/j.anbehav.2004.09.021) Andersson, M. 1994 Sexual selection. New Jersey, NJ: Gerhardt, H. C. 2005b Advertisement-call preferences in Princeton University Press. diploid–tetraploid treefrogs (Hyla chrysoscelis and Hyla Bentsen, C. L., Hunt, J., Jennions, M. D. & Brooks, R. 2006 versicolor): implications for mate choice and the evolution Complex multivariate sexual selection on male acoustic of communication systems. Evolution 59, 395–408. signaling in a wild population of Teleogryllus commodus. (doi:10.1554/04-471) Am. Nat. 167, E102–E116. (doi:10.1086/501376) Gerhardt, H. C. & Huber, F. 2002 Acoustic communication in Blows, M. W., Brooks, R. & Kraft, P. G. 2003 Exploring insects and anurans. Chicago, IL: University of Chicago complex fitness surfaces: multiple ornamentation and Press. polymorphism in male guppies. Evolution 57, 1622–1630. Gerhardt, H. C., Tanner, S. D., Corrigan, C. M. & Walton, (doi:10.1554/02-507) H. C. 2000 Female preference functions based on call Brandt, L. S. E. & Greenfield, M. D. 2004 Condition- duration in the gray tree frog (Hyla versicolor). Behav. Ecol. dependent traits and the capture of genetic variance in 11, 663–669. (doi:10.1093/beheco/11.6.663) male advertisement song. J. Evol. Biol. 17, 821–828. Henry, C. S. 1994 Singing and cryptic speciation in insects. (doi:10.1111/j.1420-9101.2004.00716.x) Trends Ecol. Evol. 9, 388–392. (doi:10.1016/0169-5347 Brodie III, E. D., Moore, A. J. & Janzen, F. 1995 Visualizing (94)90061-2) and quantifying natural selection. Trends Ecol. Evol. 10, Hosmer, D. W. & Lemeshow, S. 2000 Applied logistic 313–318. (doi:10.1016/S0169-5347(00)89117-X) regression, 2nd edn. New York, NY: Wiley.

Proc. R. Soc. B (2006) Downloaded from rspb.royalsocietypublishing.org on March 2, 2010

Female preferences and signal evolution R. L. Rodrı´guez and others 2593

Houde, A. E. & Endler, J. A. 1990 Correlated evolution of Drosophila montata courtship song and a test of the tempera- female mating preferences and male color patterns in the ture coupling hypothesis. Evolution 55,721–727.(doi:10. guppy Poecilia reticulata. Science 248, 1405–1408. 1554/0014-3820(2001)055[0721:COFPFF]2.0.CO;2) Hunt, R. E. 1994 Vibrational signals associated with mating Rodrı´guez, R. L. & Greenfield, M. D. 2003 Genetic variance behavior in the treehopper, Enchenopa binotata Say and phenotypic plasticity in a component of female mate (Hemiptera: Homoptera: Membracidae). J. NY Entomol. choice in an ultrasonic moth. Evolution 57, 1304–1313. Soc. 102, 266–270. (doi:10.1554/02-446) Hunt, J., Bussie`re, L., Jennions, M. D. & Brooks, R. 2004 Rodrı´guez, R. L., Sullivan, L. E. & Cocroft, R. B. 2004 What is genetic quality? Trends Ecol. Evol. 19, 329–333. Vibrational communication and reproductive isolation in (doi:10.1016/j.tree.2004.03.035) the Enchenopa binotata species complex of treehoppers Jang, Y. & Greenfield, M. D. 1998 Absolute versus relative (Hemiptera: Membracidae). Evolution 58, 571–578. measurements of sexual selection: assessing the contri- (doi:10.1554/03-120) butions of ultrasonic signal characters to mate attraction in Rundle, H. D. & Nosil, P.2005 Ecological speciation. Ecol. Lett. waxmoths, Achroia grisella (Lepidoptera: Pyralidae). 8, 336–352. (doi:10.1111/j.1461-0248.2004.00715.x) Evolution 52, 1383–1393. (doi:10.2307/2411308) Sattman, D. A. & Cocroft, R. B. 2003 Phenotypic plasticity Jennions, M. D., Møller, A. P. & Petrie, M. 2001 Sexually and repeatability in the mating signals of Enchenopa selected traits and adult survival: a meta-analysis. Q. Rev. treehoppers, with implications for reduced gene flow Biol. 76, 3–36. (doi:10.1086/393743) among host-shifted populations. Ethology 109, 981–994. Klappert, K. & Reinhold, K. 2003 Acoustic preference (doi:10.1046/j.1439-0310.2003.00940.x) functions and sexual selection on the male calling song in Schluter, D. 1988 Estimating the form of natural selection on the grasshopper Chorthippus biguttulus. Anim. Behav. 65, a quantitative trait. Evolution 42, 849–861. (doi:10.2307/ 225–233. (doi:10.1006/anbe.2002.2034) 2408904) Kokko, H., Brooks, R., McNamara, J. M. & Houston, A. Shaw, K. L. 2000 Interspecific genetics of mate recognition: 2002 The sexual selection continuum. Proc. R. Soc. B 269, inheritance of female acoustic preference in Hawaiian crick- 1331–1340. (doi:10.1098/rspb.2002.2020) ets. Evolution 54, 1303–1312. (doi:10.1554/0014-3820 Lin, C. P. & Wood, T. K. 2002 Molecular phylogeny (2000)054[1303:IGOMRI]2.0.CO;2) of the North American Enchenopa binotata species Shaw, K. L. & Herlihy, D. P. 2000 Acoustic preference complex (Homoptera: Membracidae). Ann. Entomol. functions and song variability in the Hawaiian cricket Soc. Am. 95, 162–171. (doi:10.1603/0013-8746(2002) Laupala cerasina. Proc. R. Soc. B 267, 577–584. (doi:10. 095[0162:MPOTNA]2.0.CO;2) 1098/rspb.2000.1040) Lorch, P. D., Proulx, S., Rowe, L. & Day, T. 2003 Condition- dependent sexual selection can accelerate adaptation. Simmons, L. W., Zuk, M. & Rotenberry, J. T. 2001 Evol. Ecol. Res. 5, 867–881. Geographic variation in female preference functions and Mead, L. S. & Arnold, S. J. 2004 Quantitative genetic models male songs of the field cricket Teleogryllus oceanicus. of sexual selection. Trends Ecol. Evol. 19, 264–271. (doi:10. Evolution 55, 1386–1394. (doi:10.1554/0014-3820 1016/j.tree.2004.03.003) (2001)055[1386:GVIFPF]2.0.CO;2) Mendelson, T. C. & Shaw, K. L. 2002 Genetic and behavioral Tregenza, T., Wedell, N. & Chapman, T. 2006 Sexual components of the cryptic species boundary between conflict: a new paradigm? Phil. Trans. R. Soc. B. 361, Laupala cerasina and L. kohalensis (Orthoptera: Gryllidae). 229–234. (doi:10.1098/rstb.2005.1796) Genetica 116, 301–310. (doi:10.1023/A:1021244812270) West-Eberhard, M. J. 1983 Sexual selection, social Olvido, A. E. & Wagner Jr, W. E. 2004 Signal components, competition, and speciation. Q. Rev. Biol. 58, 155–183. acoustic preference functions and sexual selection in a (doi:10.1086/413215) cricket. Biol. J. Linn. Soc. 83, 461–472. (doi:10.1111/ Wood, T. K. 1993 Speciation of the Enchenopa binotata j.1095-8312.2004.00404.x) complex (Insecta: Homoptera: Membracidae). In Evolu- Parri, S., Alatalo, R. V., Kotiaho, J. S., Mappes, J. & Rivero, tionary patterns and processes (ed. D. R. Lees & D. Edwards), A. 2002 Sexual selection in the wolf spider Hygrolycosa pp. 299–317. New York, NY: Academic Press. rubrofasciata: female preference for drum duration and Wood, T. K. & Guttman, S. I. 1983 Enchenopa binotata pulse rate. Behav. Ecol. 13, 615–621. (doi:10.1093/ complex: sympatric speciation? Science 220, 310–312. beheco/13.5.615) Wood, T. K. & Keese, M. C. 1990 Host–plant-induced Ritchie, M. G. 1996 The shape of female mating preferences. assortative mating in Enchenopa treehoppers. Evolution 44, Proc. Natl Acad. Sci. USA 93, 14 628–14 631. (doi:10. 619–628. (doi:10.2307/2409439) 1073/pnas.93.25.14628) Wood, T. K., Olmstead, K. L. & Guttman, S. I. 1990 Ritchie, M. G., Saarikettu, M., Livingstone, S. & Hoikkala, A. Insect phenology mediated by host–plant water relations. 2001 Characterization of female preference functions for Evoution 44, 629–636.

Proc. R. Soc. B (2006)